E-commerce CX and Returns Official

Reads a WISMO case and its order, and drafts a status reply with tracking and ETA into the case. Draft only.

1. 01Select cases to answer from the Support Cases table. If {{caseId}} is provided, work only that case. Otherwise read cases whose Type is WISMO and Status is New, up to {{maxToProcess}} (default 5). For each, find the matching order in the Orders table by Order Number and read its status, tracking, carrier, and delivery estimate.Output: For each case: the matched order and its real status, tracking, carrier, and ETA, or a note that no order matched.
2. 02Draft a friendly reply per the store voice that states the current status, tracking number, carrier, and delivery estimate from the order record. If the order is Delayed, past its ETA, Cancelled, or cannot be matched, do not improvise: set the case Status to Escalated with a short note. Otherwise put the draft in AI Draft and set Status to AI Drafted. Never invent tracking numbers or dates. Do not send anything.Output: Confirmation of the cases updated, then a short readable summary using headings and bullet points: how many drafted and how many escalated with the reason. Never output JSON or raw data.

Checks return eligibility against policy, creates a Returns and Exchanges record, and drafts a reply steering to exchange first. Draft only.

1. 01Take the return or exchange case (by {{caseId}} if provided, otherwise the newest Support Cases of Type Return or Exchange). Find the matching order in Orders. Check the store policy for the return window, condition rules, and final-sale exclusions, and judge eligibility as Eligible, Not Eligible, or Needs Review. Do not decide a denial yourself; mark genuine denials as Needs Review for a human.Output: The eligibility judgement with the policy basis, and the matched order details.
2. 02Create a row in the Returns and Exchanges table with the RMA details, reason, eligibility, a suggested resolution (prefer Exchange or Store Credit for size and fit reasons), and Status Requested. Do not set Label Sent.Output: Confirmation of the Returns row created with its eligibility and suggested resolution.
3. 03Draft a friendly reply per the store voice: if eligible, explain the options and gently steer toward an exchange or store credit before a refund; if it needs review, say it is being looked into rather than promising an outcome. Put the draft in the case AI Draft and set the case Status to AI Drafted, or Escalated if it needs human review. Never promise a refund above the approval threshold or a denial on your own. Do not send anything.Output: Confirmation the case was updated, then the readable reply draft and the eligibility outcome. Never output JSON or code blocks.

Confirms refund amount from order and return records, drafts a confirmation, and flags over-threshold refunds for human approval. Draft only.

1. 01Look up the return {{rmaId}} in the Returns and Exchanges table and its order in Orders. Confirm the correct refund amount from the order total and policy (for example, whether return shipping or a restocking fee applies). Check the policy's refund approval threshold.Output: The confirmed refund amount with how it was calculated, and whether it is over the approval threshold.
2. 02Record the refund amount and set the Returns row Resolution to Refund and Status to Approved only if it is within the threshold; if it is over the threshold, leave Status as Requested and mark it for human approval. Draft a refund confirmation message per the store voice that does not promise timing beyond policy. Never process or send the refund.Output: A readable result with clear labels: refund amount and basis, approval status (ready or needs human approval), and the confirmation draft. Never output JSON or code blocks.

Drafts short review or feedback requests for recently resolved, satisfied customers. Draft only.

1. 01Read recently Resolved cases from the Support Cases table whose CSAT is Positive or Neutral (or unrated but clearly resolved well), up to {{maxToProcess}} (default 10). Skip anything with Negative CSAT, a complaint, a refund dispute, or an escalation.Output: The list of satisfied resolved cases eligible for a review request.
2. 02For each, draft a short, warm review or feedback request per the store voice, referencing the resolution lightly and making it easy to leave a review. Put the draft in the case AI Draft or Resolution Notes. Do not ask dissatisfied customers and do not send anything.Output: Confirmation of the cases updated, then a short readable summary using headings and bullet points: how many review requests were drafted. Never output JSON or raw data.

Reads the CX tables and produces a daily digest of volume, deflection, top case types, and return and complaint drivers.

1. 01Read recent rows from Support Cases and Returns and Exchanges. Compute case volume, the split of AI Drafted versus Escalated versus Resolved, an approximate deflection read, the top case types, the most common return reasons, and any complaint or damage trend.Output: The computed totals and top lists from the tables.
2. 02Write a concise daily CX digest for the owner. Optionally record the day's totals as a row in CX Metrics.Output: A readable digest using headings and bullet points: volume and deflection, top case types, return reasons, and anything to watch. Lead with the headline numbers. Never output JSON or raw rows.